|
On the attached logs we see a DDL event succeeding in two of the nodes but failing in the third one at 2:31:57:
2018-06-07 2:31:57 140668603120384 [Note] WSREP: abort transaction: BF: /* ApplicationName=DBeaver 2.4.1 - Main connection */ *alter table delivery add column lock_count integer* victim: (null)
|
2018-06-07 2:31:57 140668603120384 [Note] WSREP: cluster conflict due to high priority abort for threads:
|
The DDL was retried at 3:44:47 and failed again, finally succeeding at 13:07:58:
2018-06-07 13:07:58 140186596428544 [Note] WSREP: TO BEGIN: -1, 0 : alter table appt_scheduler_PROD.delivery add column lock_count integer
|
2018-06-07 13:07:58 140186596428544 [Note] WSREP: TO BEGIN: 27163806, 2
|
2018-06-07 13:07:58 140186596428544 [Note] InnoDB: Online DDL : Start
|
(...)
|
2018-06-07 13:08:00 140186596428544 [Note] InnoDB: *Online DDL : Completed*
|
2018-06-07 13:08:00 140186596428544 [Note] WSREP: avoiding binlog rotate due to TO isolation: 1
|
It should not have completed on the other two nodes if one of the nodes was unable to enter Total Order Isolation.
|